SAM, PVWATTSv1 and PVWATTSv5 discrepancies

Hi,

I'm doing some simple modelling of PV Systems using PVWATTS, but getting big discrepancies in my results. I wonder if you can help?

I have been using the PVWATTSv1 module in the SDK, called from within MATLAB to model a PV system for a year. Today i wanted to check my results so i ran the same data using the PVWATTSv5 module, and again using the PVWATTS option through the SAM user interface (Version 2015.6.30, 64 bit, updated to revision 4).

I'm using the same weatherfile (attached) for every run.

I used 80% degradation for v1 and the default 14% losses for v5 and for SAM, which should be equivalent (by my calculations).

I have also attached the hourly output for each method, and a screenshot showing the inputs. (The matlab system size, tilt and azimuth aren't shown, but are set to 47.25, 2degrees and 344 degrees respectively).

The ac total output is 5% higher using SAM than v1.
But the output is 21% higher using v5 than v1,

I am surprised if such a big discrepancy is due to differences in the algorithm, so i am pretty sure i am doing something stupid - but for which method??

Appreciate any help you can give me.

Thanks,

Mike

Dear Mike,
First of all, for a description of PVWatts V5 and the differences between V1 and V5, please see Dobos (2014) listed under PVWatts on the SAM Performance Model Documentation page of this website. Differences between the two models are highlighted in boxes throughout the document.
In your inputs to the pvwattsv5 module, the losses input should be 14, not 0.14. 
Also, to be more consistent with pvwattsv1, set module_type = 0 (standard) and array_type = 0 (fixed open rack).
Please see Section 11 of the PVWatts V5 reference manual for a description of how to convert between a V1 derate factor and V5 loss value.
Because of the differences between the V5 and V1 models, you should see some difference in the results, even after you calibrate the inputs to be as similar as possible. But, the differences should be relatively small. V5 is the current version available on the PVWatts website and that SAM uses, so you should consider that to be the more accurate model.
Best regards,

Paul.

Dear Paul,

Thanks for getting back so quickly. That's brilliant - I still have 5% discrepancy between v1 and v5. But v5 with the parameters you've given me is within .1% of the result running the SAM UI.

Many thanks for your help.

Mike